The Times They Were A-Changin
Transcription
The Times They Were A-Changin
Empirical Musicology Review Vol. 10, No. 3, 2015 “The Times They Were A-Changin’”: A Database-Driven Approach to the Evolution of Harmonic Syntax in Popular Music from the 1960s HUBERT LÉVEILLÉ GAUVIN [1] Schulich School of Music, McGill University ABSTRACT: The goal of this research is to investigate the pitch structures of popular music in the 1960s through a large corpus study in order to identify any consistent changes in harmonic and tonal syntax. More specifically, two studies based on the Billboard DataSet (Burgoyne, Wild & Fujinaga, 2011; Burgoyne, 2011), a new corpus presenting transcriptions for more than 700 songs, are presented. The first study looks at the incidence of multi-tonic songs throughout the decade, while the second study focuses on the incidence of flat-side harmonies (e.g. bIII, bVI, and bVII) over the same period of time. While no difference was observed in the frequency of multi-tonic songs, the study showed a significant increase in the incidence of flat-side harmonies during the second half of the decade. Submitted 2014 September 16; accepted 2015 April 9. KEYWORDS: popular music, corpus, Billboard, harmony, modulation IN January 1964 Bob Dylan released “The Times They Are A-Changin,” a politically-charged protest song encouraging changes in American society. Indeed the 1960s were a time of great political, but also sociological and cultural changes, accompanied by equally important developments in popular music. Many music scholars have addressed the attitudinal shift associated with this period, noting that “[r]ock musicians no longer aspire[d] so much to be professionals and craftspeople” but “artists” (Covach, 2006, p. 38) and that the “later 1960s brought a respect for popular music and a popularity for complex artistic experimentation that had not been matched in any previous era” (Wald, 2009, p. 246). In recent years, theorists have highlighted specific paradigms of pitch structures in popular music, especially associated with rock. Their observations include an increasing use of modal harmonies and modally-derived chord progressions (Moore, 1992, 1995; Biamonte, 2010), plagal progressions (Temperley, 2011) and an increasing use of chord loops (Tagg, 2009). While these authors seem to agree, at least partially, on some harmonic tendencies that characterize rock music, it is difficult to relate these new idioms to the early pop and rock ‘n’ roll music that predates the aforementioned shift. Even though attempts have been made in the past to empirically map these progressions onto a specific timeline (Everett, 2004; de Clercq & Temperley, 2011, 2013), no previous research has focused on change within a specific decade. The goal of this research is to investigate pitch-based structures in this repertoire through a large corpus study in order to identify any consistent changes in harmonic and tonal syntax. The Billboard DataSet (Burgoyne, Wild & Fujinaga, 2011; Burgoyne, 2011) provides a collection of 743 transcriptions of music popular in the United States between 1958 and 1991.[2] The corpus was originally created to “enable significant advances in the quality of training for audio-chord-recognition algorithms” as well as to engage in “computational musicology” (Burgoyne, et al., 2011, p. 633). It consists of a random sampling from all the songs that made the weekly Billboard 100 charts throughout this time period. The content of the corpus is multigenre, primary consisting of rock ‘n’ roll, pop/rock, r ‘n’ b/soul, coutry & western, vocal, and folk material. Since Billboard is an American standard record chart, most of the songs come from the U.S.A, the U.K., and Canada, though other countries can be represented as well. The transcribing team consisted of more than two dozen people, all university-trained jazz musicians. Each selected song was annotated separately by two different transcribers, then the analyses were reconciled by a third. Empirically oriented surveys of popular music are becoming more frequent. Perhaps the most important study for the field in recent years is that by de Clercq and Temperley (2011). They presented a 100-song corpus based on Rolling Stone Magazine’s “500 Greatest Songs of All Time,” compiled in 2004.[3] Their study focused primarily on chord frequency, frequency of root motions, and patterns of co-occurrence between chords. Although the Billboard DataSet focuses on popular music from the same time period as the 215 Empirical Musicology Review Vol. 10, No. 3, 2015 Rolling Stone corpus, its size is significantly larger. In fact, it is, to my knowledge, the largest collection of methodologically curated transcribed popular music annotations. Furthermore, it focuses on songs that were considered popular in their respective time, instead of songs that were considered important in retrospect. As such, it appeared to be the best available tool for the present research. Two different aspects of the above-mentioned shift were tested in two different studies, the first dealing with modulation and the second with chord frequency. Since the primary focus of this research is the evolution of harmonic syntax throughout the 1960s, the time period under consideration is 1958 to 1971. These 14 years constitute a sub-corpus of 292 songs. A detailed list of the songs featured in this sub-corpus can be found in the Appendix. The first study focuses on modulation in the 1960s. Different scholars have discussed modulatory devices in popular music, especially in rock. Everett (1997) mentioned how the “truck driver’s” modulation by semitone can be motivated by various reasons, such as “signaling transcendence in a story line,” “portray[ing] […] the passage of time,” or “provid[ing] a change of colour for the ‘big finish.’” (p. 151). Similarly, Osborn (2013) discussed how “experimental rock artists regularly end songs with completely new material designed to be more memorable than anything previously presented—the terminal climax” and how these “dramatic endings come about through any combination of amplitudinal climax, harmonic modulation, and changing meter. ” (p. 23). Alternatively, Tagg (2009) noted how dominant modulations are so “indicative of European art music that they can be inserted as genre synecdoches in a context of non-‐classical harmony (e.g. pop and rock) to connote, seriously or humorously, high art rather than low-brow entertainment, deep feelings and the transcendent rather than the superficial and ephemeral” (p. 111). Considering the premise introduced earlier in this paper that songwriters tried to move from craftspeople to artists during this time period, it is reasonable to think that different devices might have been experimented with to achieve this goal, including modulation. As such, the proposed theory for this first study is that the incidence of songs featuring modulations increased through the decade. The second study focuses on the frequency of chords bIII, bVI, and bVII[4], the so-called “flat-side” harmonies. As Everett (2001) mentions, “[f]lat-side scale degrees appear primarily within the minor key […] and through mode mixture in the major key.” (p. 53). De Clercq and Temperley (2011) investigated chord frequency, frequency of root motions, patterns of co-occurrence between chords, and melodic organization in popular music. Their findings include a dramatic shift between chords used during the 1950s and the 1960s to 2000s, the 1950s being “completely dominated” (p. 63) by I, IV, and V. Conversely, the authors discussed how, in the 1960s onward, flat-side harmonies bVII, bIII, and bVI “emerge as a group in which all three pairs are highly correlated” (p. 66). These findings match the modal characteristics associated with rock (Moore, 1992, 1995; Everett, 2004; Biamonte, 2010), a genre that became very popular during the late 1960s. Taking into consideration this shift between the 1950s and the later decades, the proposed theory for this second study is that, as we go further into the decade, chords bIII, bVI and bVII increased in frequency. HYPOTHESES Formally, the hypothesis for the first study is: H1 Songs featuring more than one tonal center will increase in frequency over the studied time period of 1958-1971. For the second study, the formal hypothesis is: H2 Songs featuring flat-side harmonies (i.e. bIII, bVI, and bVII) will increase in frequency over the studied time period of 1958-1971. To anticipate the conclusions, the results were inconsistent with the first hypothesis, but consistent with the second. METHOD Parsing the Database The present study relies on the Billboard DataSet, which presents a corpus of harmonic transcriptions for 743 different songs, including 292 in the studied period of 1958-1971. The chronological distribution of the 292 songs is presented in Table 1. 216 Empirical Musicology Review Vol. 10, No. 3, 2015 Table 1: Distribution by year of Songs in the Billboard DataSet for 1958-1971. Year 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 Number of Transcribed Songs 7 14 9 20 27 20 25 20 18 28 26 24 22 32 The transcriptions were created in plain-text format. Each file starts with a header that includes meta-information related to each song: title, artist, meter, and key. Those lines are preceded by the comment character hash ( # ) to distinguish them from the actual annotation. The notational system used is based on a standardized approach (Harte et al., 2005), with vertical slashes ( | ) used to represent barlines. Each individual line of annotation is preceded by the timestamp of the beginning of the phrase, expressed in seconds. Annotators could freely add other information before or after the vertical slashes, such as form, instrumentation, etc. This transcription format proved to be polyvalent yet created some obstacles. In order to rapidly and automatically parse through the large number of transcriptions, a UNIX tool was created. This script allows the user to use regular expression (regex, a character string used for patternmatching) to establish proper queries. However, since the files feature extra information such as audio timing and annotators’ comments, reading a file as a long single line would fail to have the progressions adjacent to one another. Therefore, the files were processed to remove all inessential information: in every line that started with a comment character ( # ) as well as every blank line was kept exactly the same, while in all other lines any characters not enclosed by vertical slashes were removed. Other problems were linked to using the script. Since the command searches strings of characters, it was impossible in this transcription format to search for a single progression across songs in different keys. Indeed, when dealing with strings of characters, “D:min7 G:7” in C is different from “C:min7 F:7” in Bb, even though both progressions bear the same relation with their respective tonal center. One way to overcome this problem is to convert the original transcriptions into a tonic-neutral format, where the root of each chord is replaced with integer notation (where t and e stand for 10 and 11, respectively). The conversion was done automatically with a second script that relied on custom dictionaries for every possible tonic, including enharmonic equivalence. The script would read the tonic of each song, as notated in the header, decide which dictionary to use, and then convert every root to its neutral equivalent. Songs featuring more than one tonal center (69 in total) were dealt with manually—separated into single-tonic sections, converted using the same procedure, and then reassembled. The newly formatted files were then saved using the same name as their original counterpart, but with a different extension. The new files kept the information in the same order as the original files, allowing the user to easily go back and forth between this format and the original transcriptions. Figures 1.1 and 1.2 show an original transcription and a reformatted one for comparison. 217 Empirical Musicology Review Vol. 10, No. 3, 2015 Fig. 1.1. An excerpt of a transcription as presented originally in the Billboard DataSet. Fig. 1.2. The transcription shown in Fig. 1.1 converted to a tonic-neutral format and cleaned of all nonessential information. Subdividing the Corpus In order to identify any significant changes in harmonic syntax throughout the studied time period, the 14-year time span was divided into two subspans of seven years each: the early 60s (1958-64) and the late 60s (1965-71). Those two subspans contain 122 and 170 songs, respectively. This important difference between the two subspans is due to two main factors. First, the Billboard Hot 100 charts (and thus the database) started in August 1958, whereas all the other years start in January. This explains why the year 1958 features only 7 songs. Second, the random sampling procedure used by Burgoyne et al. created a discrepancy between the different years. Figure 2 reprints Figure 1 from Burgoyne et al. (2011, p. 634) and explains in detail the sampling algorithm used to create the database. While most years in the 14-year period under study feature between 20 and 30 songs, 1959 218 Empirical Musicology Review Vol. 10, No. 3, 2015 and 1960 are outliers with 14 and 9 songs, respectively. Conversely, 1971 features 32 songs, which is higher than average. This important discrepancy between the two subspans had an impact on the methodology used to analyze the data. Originally, a methodology was designed which aimed to look for gradual linear changes across the whole 14-year period. However, considering that the number of songs available for each year varies substantially, the results would have been distorted by this approach. Instead, a methodology was used which compared the early 60s with the late 60s, avoiding oversampling any single year due to a larger number of songs in the database for that year, and instead considered broader changes between the beginning and the end of the decade. Fig. 2. Sampling algorithm for the Billboard DataSet. Reproduced from Burgoyne et al. (2011). 219 Empirical Musicology Review Vol. 10, No. 3, 2015 RESULTS Study 1: Evaluating Modulation Frequency The first study focused on modulation. In order to find the number of modulating songs, every file featuring the character string “tonic:” more than once was counted. For example, Jan & Dean’s Sidewalk Surfin’, previously presented in Figures 1.1 and 1.2, features two tonics, D and Eb, and so would qualify as a modulating song. Out of the 292 songs searched, 31 multi-tonic songs were found. Figures 3.1. and 3.2 present the distribution of the results. 100% Distribution of Single- and Multi-Tonic Songs Results 80% 60% Single-Tonic Songs 40% Multi-Tonic Songs 20% 0% 1958-64 1965-71 Fig. 3.1. Distribution of single- and multi-tonic songs in the Billboard DataSet between 1958 and 1971, arranged in five-year bins. 25% Results 20% 15% Multi-Tonic Songs 10% 5% 0% 1958 1960 1962 1964 1966 1968 1970 Fig. 3.2. Frequency of multi-tonic songs in the Billboard DataSet between 1958 and 1971, plotted by year. Using a chi-squared test at a confidence level established at 99%, the difference between the number of multi-tonic songs found in 1958-64 and 1965-71 was not statistically significant (χ2= 3.78; df=1; p=.05). As such, there appears to be no significant change in the frequency of multi-tonic songs over time. Study 2: Evaluating Chord Frequency The second study focused on the frequency of flat-side harmonies over time. For the sake of comparison, the frequency of 24 different chords was considered (12 chromatic roots, with either 220 Empirical Musicology Review Vol. 10, No. 3, 2015 major or minor triads). Table 2 shows the overall proportion of songs featuring those chords in each subcategory (i.e 1958-64 and 1965-71). The results for each subcategory were then compared with one another, again using a chi-squared test with a confidence level of 99%. Note that no significance test was done on chords featuring in less than 5% of the corpus (i.e. bii, biii, #IV, #iv, bvi, bvii, VII, vii), as the data collected for those chords was deemed to small to be representative. Due to the way truckdriver’s modulations were notated in the Billboard DataSet, many chords immediately preceding a new tonic were originally notated as bVI in the original key, but acted as a tonicization of the tonic in the following key (e.g. V-I in the new key). To address this notational problem, multi-tonic songs featuring at least one bVI chord were manually verified. Songs where the only bVI chord present was immediately before a truck-driver’s modulation were not taken into account in the following table. Admittedly, this creates a bias in the data. However, considering that there is no significant difference in the distribution of multi-tonic songs between the early and late 60s, the bias created appears to be less intrusive than the one created by the original notational process. The results are summarized in the third column of Table 2. Starred (*) p values indicate statistical significance at the 99% confidence level. Table 2: Distribution of songs featuring specific chords (in percentage). I i bII bii II ii bIII biii III iii IV iv #IV #iv V v bVI bvi VI vi bVII bvii VII vii Early 60s (1958-64) 98.36% 5.06% 8.2% 0% 37.71% 33.61% 10.66% 2.46% 19.67% 19.67% 89.34% 13.14% 4.91% 1.64% 96.72% 5.74% 9.84% 0% 17.21% 47.54% 13.93% 0.82% 4.92% 0.82% Late 60s (1965-71) 85.29% 22.36% 5.88% 0.59% 29.41% 35.29% 28.24% 4.92% 18.82% 20.59% 83.52% 21.18% 2.36% 0% 80.59% 15.88% 21.77% 1.18% 18.24% 39.41% 39.41% 2.35% 3.52% 2.29% Results χ2=14.45; df=1; p<.01* χ2=15.04; df=1; p<.01* χ2<.01; df=1; p=.96 N/A χ 2=2.21; df=1; p=.14 χ2=0.09; df=1; p=.77 χ2=13.28; df=1; p<.01* N/A χ2=.03; df=1; p=.86 χ2=.04; df=1; p=.85 χ2=1.99; df=1; p=.16 χ2=2.51; df=1; p=.11 N/A N/A χ2=16.71; df=1; p<.01* χ2=7.11; df=1; p=.01* χ2=7.24; df=1; p<.01* N/A χ2=.05; df=1; p=.82 χ2=1.92; df=1; p=.17 χ2=22.5; df=1; p<.01* N/A N/A N/A The results presented above are consistent with the proposed hypothesis that songs featuring flat-side harmonies (i.e. bIII, bVI, bVII) increase in frequency over the studied time period. DISCUSSION The hypothesis for the first study arose from a simple premise: in order to move from craftspeople to artists, songwriters may have experimented with different compositional devices. As such, it appeared reasonable to question whether modulation was one such device. However, as the results of Study 1 showed, no significant change occurred between the early 60s and the late 60s. Looking at the results in more detail, three different types of modulations can be distinguished. The first type, though sometimes bearing different names, is the truck driver’s modulation: “a sudden shift from one tonal center to another—usually a half step [but sometimes a full step]—that is not functionally related to the first.” (Everett, 1997, p. 118, comment in brackets mine). A second type of modulation can be described as a sectional modulation, where a specific section of a song is in a different key from the rest of the piece. With this type of modulation, a song begins and ends in the same tonality. A third type occurs when a song changes key permanently, but with a modulation that is 221 Empirical Musicology Review Vol. 10, No. 3, 2015 not a semitone or a tone. The distribution of these three types of modulation over the 31 multi-tonic songs found between 1958 and 1971 is presented in Figure Distribution of Multi-Tonic Songs, per Type 4. of Modulation Truck Driver’s Modulation Other Sectional Modulation Fig. 4. Distribution of 31 multi-tonic songs between 1958-1971: Truck Driver’s Modulation (28 songs, 90.32%); Sectional Modulation (2 songs, 6.45%); Other (1 song, 3.23%). As shown in Figure 4, the truck driver’s modulation is the most frequently used modulation technique between 1958 and 1971. The majority of the songs (23 songs, 82.14%) featuring this type of modulation only modulate once, while songs modulating two (3 songs, 10.71%) or three times (2 songs, 7.14%) are far less frequent. It also appears to be very infrequent for multi-tonic songs with truck driver’s modulation to feature both modulations by semitone and by tone (1 song, 3.57%). Although the hypothesis for the second study focused on only flat-side harmonies, the frequency of songs featuring 24 different chords was analyzed. As hypothesized, bIII, bVI, and bVII occurred in a significantly greater number of songs in the second half of the decade. A decrease was observed of roughly 15 % in songs featuring major tonic chords (i.e. I), matched by a similar increase in songs featuring minor tonic chords (i.e. i); moreover, minor dominant chords (i.e. v) also increased in frequency in the later 60s. Although it is not possible at this point to conclude that there is a correlation between both results, it would be worthwhile to follow this lead in future work and investigate whether minor songs become statistically more frequent in the second half of the decade. Investigating Chord Patterns Considering that bIII, bVI, and bVII became more prominent during the second half of the decade, it would be interesting to examine the harmonic context in which they were used. In their study using the Rolling Stone corpus, de Clercq and Temperley (2011) investigated the “frequency of relative-root ‘trigrams,’” using this term to describe “groups of three adjacent chords” (p. 63). Table 3, reprinted from de Clercq and Temperley (2011), shows the most frequent harmonic trigrams ending with a tonic chord, for the 1950s through to the 1990s. Table 3: Distribution of songs from de Clerq and Temperley’s (2011) Rolling Stone corpus with harmonic trigrams ending in the tonic (and not beginning in tonic), in descending order of frequency (reproduced from de Clercq and Temperley, 2011, Table 7). Trigram Instances IV V I 352 V IV I 292 bVII IV I 146 VI IV I 126 bVII bVI I 103 bIII bVI I 66 II V I 63 bVI bVII I 60 V VI I 42 IV bVII I 39 222 Empirical Musicology Review Vol. 10, No. 3, 2015 A similar approach was taken to observe the harmonic syntax involving flat-side harmonies. Using the chords presented in Table 2, every possible permutation featuring at least bIII, bVI, or bVII, and ending with a tonic chord was used to parse the database. However, the methodology differed from the one used by de Clercq and Temperley in two ways: chord mode (major or minor) was taken into account instead of only relative-root motion, and the number of songs featuring a specific progression was calculated, instead of instances of said progression. Chords could occur as simple triads or ornamented with extensions (e.g. maj6, maj7, add9). Moreover, major triads could also occur as dominant seventh or dominant ninth chords. Open-fifth chords (“power chords”) were not taken into consideration. The parser was designed to look for continuous strings of characters; as such the search mechanism avoided progressions running over modulations, as both tonalities in the transcription file were separated by a comment line indicating the new key, as shown in Figure 1.2 (e.g. #tonic: Eb). For this reason, the line preceding the comment line may display some odd progressions, as the last chord of such lines is usually used to modulate to the new key. However, since the number of songs featuring a specific progression was taken into account, and not the specific instances of a progression, this potential distortion did not affect the parsing process. The results for the 1958-72 period are presented in Table 4. Table 4: Distribution of songs from the Billboard DataSet with harmonic trigrams ending in tonic (and not beginning in tonic) and featuring at least one flat-side harmony, in descending order of frequency. Trigram bVII IV I bVII V I bIII IV I bVI V i bIII V I bVI V I Early 60s (1958-64) 0 4 1 0 1 3 Late 60s (1965-71) 14 8 4 5 3 1 Total 14 12 5 5 4 4 The table presented above gives us a more detailed account of how flat-side harmonies were used during the decade. Although it has been established previously that all three flat-side chords (i.e. bIII, bVI, bVII) increased significantly in frequency throughout the decade, there is a clear preference for the bVII-IV-I trigram. This progression is often referred to as a double plagal progression, as the movement from bVII to IV imitates the plagal movement going from IV to I, as Everett explains: In the double plagal progression a chain of descending fourth emerges, with a major IV of IV created by lowering the root of a vii chord to b7, creating the bVIIIV-I motion […] In this chord succession, bVII resolves to IV with a transposed version of the same descending neighbour motions (here 2-1 and b7-6) used by IV in resolving to I (6-5 and 4-3). Originally appearing in 1957-60 as an ornamental guitar figure […], this function emerged in broader rhythms in the soul music of late 1964 […] and early 1965 […] to become rock mainstay. (2009, p. 274). Everett’s comments match very closely the results of this study, with the emergence of the double plagal progression during the late 60s. His description also exemplifies the modal aspect of the rock genre, suggesting a mixolydian sound created through the flattening of scale degree 7. The plagal direction of this progression also matches Stephenson’s (2001) theory that, “[w]ith regard to harmonic succession, as with cadence placement, rock has, from its beginning, used a style opposed to that of common practice, a style that became increasingly standard during the late 1950s and the 1960s” (p. 103). Other theorists have seen tonal processes in rock, despite its modal characteristics. For example, Biamonte (2010) has described how bVII-IV-I, along with the Aeolian progression bVIbVII-I, tends to behave tonally: “Although their pitch syntax is not diatonically tonal, in many cases these two characteristic chord patterns express traditional tonal functions such as tonic prolongation and dominant preparation, both of which are often accomplished through elaboration of the tonic by subdominant harmony and cadential resolution.” (p. 98).[5] This dual nature of the progression, a modal sound behaving tonally, may explain, at least partially, why this specific pattern was significantly favored over other progressions featuring flat-side harmonies, as it serves as a bridge between common and new practices. The second most frequent trigram, bVII-V-I, is a modally-mixed progression particularly noticeable for its cross-relation between the flat and natural scale degree seven. This type of crossrelation occurs frequently in progressions using flat-side harmonies, and mostly results from guitar223 Empirical Musicology Review Vol. 10, No. 3, 2015 oriented gestures, using parallel barre chords up and down the fretboard. In an article discussing the topography of the guitar and its impact on pop-rock music, Koozin (2011) showed how “physical constraints inherent in guitar playing may shape musical material” (par. 23). Indeed, the same phenomenon happens with bIII-IV-I, bIII-V-I, and bVI-V-I. This reflects the idea that throughout the 1960s, music shifted from being primarily composed on the keyboard to being primarily composed on the guitar. This new practice became heavily associated with rock, so much so that by the late 1970s, “pop” referred to keyboard-based music while “rock” referred to guitar-based music. (Nobile, 2014, p. 8). CONCLUSION The goal of carrying out the studies presented in this paper was to test whether significant changes in harmonic practices during the 1960s matched the attitudinal shift discussed by popular music scholars. Two studies were conducted, the first focusing on modulating songs, the second focusing on flat-side harmonies. Although the results showed no significant difference between the use of modulation in the first and second half of the decade, there was a significant increase in frequency of flat-side harmonies in the second half of the decade, as hypothesized. However, it would be misleading to assume that this new genre, rock, and its new harmonic practices took over the entire popular music world. Though a new harmonic practice seems to be emerging during the studied period, it does not appear to replace more traditional diatonic harmony, but rather coexists with it. For example, the results presented in Table 4 show bVII-IV-I as the most frequent progression using a flat-side harmony, but the frequency of this progression is still relatively marginal, with only 14 songs out of 170 (8.24%) featuring this progression. By comparison, IV-V-I and V-IV-I, the two most popular diatonic trigrams identified by de Clercq and Temperley in Table 3, were featured in 91 (53.53%) and 50 (29.41%) songs, respectively. Furthermore, although flat-side harmonies increased in frequency through the decade, they did not do so through clearly established idiomatic progressions, bVII-IV-I and, to a lesser extent, bVII-V-I being exceptions. What may be misleading, then, is the tendency to consider harmonic progressions or songs that were historically significant the norm of a certain time period. Lists such as Rolling Stone’s “500 Greatest Songs of All Time,” used by de Clercq and Temperley (2011; 2013), are useful to study characteristic tendencies of a specific genre, but fail to give a clear image of a time period as they present songs that were considered historically important a posteriori, and as such, are biased towards innovative or long-lasting, successful songs. When looking at the results presented here, based on a corpus of songs deemed popular in their own time, the attitudinal shift discussed by some scholars appears to be matched with new harmonic practices. However, those new practices established themselves over time. The two studies presented here focused on very specific aspects of harmonic syntax: modulation and flat-side harmonies. Yet, in order to have a better understanding of the evolution of musical tendencies, other aspects would benefit from a similar empirical approach. Broadening the queries to include progressions that do not necessarily feature tonic chords would help to provide a clearer image of harmonic practices. Similarly, secondary parameters such as hypermeter, form[6], rhythm, and timbre, should be taken into consideration. Although some scholars have looked empirically at those parameters (see Huron & Ommen, 2006; Serrà, et al., 2012; Biamonte, 2014), they still remain largely under-researched. ACKNOWLEDGMENTS This research project was supported by grants from the Social Sciences and Humanities Research Council of Canada and the Fonds de recherche du Québec - Société et culture. Special thanks go to Nicole Biamonte and Jonathan Wild for their valuable help. NOTES [1] Email correspondence can [email protected] be addressed to Hubert Léveillé Gauvin at [2] Burgoyne et al. explain the basis of the chronological span of this corpus as follows: “The date of the first chart, 4 August 1958, is a natural starting date for selecting songs, but choosing an end date is less straightforward. Hip-hop music does not lend itself readily to harmonic analysis as traditionally understood, and because hip-hop became more popular in the 1990s and 2000s, a larger portion of the music on the ‘Hot 100’ chart from these periods falls out of the scope of the data set. Furthermore, there have been several changes to the formula for computing the ‘Hot 100’ over time, including a particularly significant shift in December 1991 […]” (Burgoyne, et al., 2011, p. 634) 224 Empirical Musicology Review Vol. 10, No. 3, 2015 [3] Since the publication of their article in 2011, de Clercq and Temperley have expanded their corpus to 200 songs (Temperley & de Clercq, 2013). [4] Throughout this paper, roman numeral are shown in relation to major, an approach used by Biamonte (2010), Temperley (2011), de Clercq (2012), Nobile (2014), among many others. [5] For a more thorough discussion on double plagal progressions, see Everett, 2001, 2004; Carter, 2005; Spicer, 2005; Biamonte, 2010. [6] Although the annotators working on the Billboard DataSet were encouraged to “use free comments in particular to denote major structural features such as verses, bridges, and choruses” (Burgoyne, 2011, p. 194), those formal labels were not properly curated, which make their use obsolete for anything other then informal discussion. REFERENCES Biamonte, N. (2010). Triadic modal and pentatonic patterns in rock music. Music Theory Spectrum, 32(2), 95–110. Burgoyne, J. A. (2011). Stochastic processes and database-driven musicology. (Unpublished doctoral dissertation). McGill University, Montreal. Burgoyne, J.A., Wild, J., & Fujinaga, I. (2011). An expert ground-truth set for audio chord recognition and music analysis. In A. Klapuri & C. Leider (Eds.), Proceedings of the 12th International Society for Music Information Retrieval Conference (pp. 633-638). Miami, Florida: University of Miami. Carter, P. S. (2005). Retrogressive harmonic motion as structural and stylistic characteristic of poprock music. (Unpublished doctoral dissertation). University of Cincinnati, Cincinnati. Covach, J. (2006). From ‘craft’ to ‘art’: Formal structure in the music of the Beatles. In K. Womack & T. F. Davis (Eds.), Reading the Beatles: Cultural Studies, Literary Criticism, and the Fab Four (pp. 37-53). Albany: State University of New York Press. de Clercq, T. (2012). Sections and successions in successful songs: A prototype approach to form in rock music. (Unpublished doctoral dissertation). Eastman School of Music, Rochester. de Clercq, T., & Temperley, D. (2011). A corpus analysis of rock harmony. Popular Music, 30(1), 47– 70. Everett, W. (1997). Swallowed by a song: Paul Simon's crisis of chromaticism. In J. Covach & G. M. Boone (Eds.), Understanding Rock: Essays in Musical Analysis (pp. 113-58). Oxford: Oxford University Press. Everett, W. (2001). The Beatles as Musicians: The Quarry Men through Rubber Soul. Oxford: Oxford University Press. Everett, W. (2004). Making sense of rock's tonal systems. Music Theory Online, 10(4). Retrieved from http://www.mtosmt.org/issues/mto.04.10.4/mto.04.10.4.w_everett.html Everett, W. (2009). Foundations of Rock: From ‘Blue Suede Shoes’ to ‘Suite: Judy Blue Eyes.’ Oxford: Oxford University Press. Harte, C. A., Sandler, M. B., Abdallah, S. A. & Gómez, E. (2005). Symbolic representation of musical chords: A proposed syntax for text annotations. In Proceedings of the 6th International Conference on Music Information Retrieval (pp. 66-71). Retrieved from http://ismir2005.ismir.net/proceedings/1080.pdf Huron, D., & Ommen, A. (2006). An empirical study of syncopation in American popular music, 1890–1939. Music Theory Spectrum, 28(2), 211-31. Koozin, T. (2011). Guitar voicing in pop-rock music: A performance-based analytical approach. Music Theory Online, 17(3). Retrieved from http://www.mtosmt.org/issues/mto.11.17.3/mto.11.17.3.koozin.php 225 Empirical Musicology Review Vol. 10, No. 3, 2015 Moore, A. (1992). Patterns of harmony. Popular Music, 11(1), 73-106. Moore, A. (1995). The so-called ‘flattened seventh’ in rock. Popular Music, 14(2), 185-201. Nobile, D. F. (2014). A structural approach to the analysis of rock music. (Unpublished doctoral dissertation). The City University of New York, New York. Osborn, B. (2013). Subverting the verse/chorus paradigm: Terminally climatic forms in recent rock music. Music Theory Spectrum, 3(1), 23-47. Serrà, J., Corral, Á., Boguñá, M., Haro, M., & Arcos, J. L. (2012). Measuring the evolution of contemporary Western popular music. Scientific Reports, 2. Retrieved from http://www.nature.com/srep/2012/120726/srep00521/full/srep00521.html Stephenson, K. (2002). What to Listen for in Rock: A Stylistic Analysis. New Haven, CT: Yale University Press. Spicer, M. (2004). (Ac)cumulative form in pop-rock music. Twentieth Century Music, 1(1), 29-64. Spicer, M. (2004). Review of Walter Everett, The Beatles as Musicians: The Quarry Men Through Rubber Soul. Music Theory Online, 11(4). Retrieved from http://www.mtosmt.org/issues/mto.05.11.4/mto.05.11.4.spicer.html Summach, J. (2011). The structure, function, and genesis of the prechorus. Music Theory Online, 17(3). Retrieved from http://www.mtosmt.org/issues/mto.11.17.3/mto.11.17.3.summach.html Summach, J. (2012). Form in top-20 rock music, 1955–89. (Unpublished doctoral dissertation). Yale University, New Haven. Tagg, P. (2009). Everyday Tonality: Towards a Tonal Theory of What Most People Hear. New York: Mass Media Scholar's Press. Temperley, D. (2011). The cadential IV in rock. Music Theory Online, 17(1). Retrieved from http://www.mtosmt.org/issues/mto.11.17.1/mto.11.17.1.temperley.html Temperley, D., & de Clercq, T. (2013). Statistical analysis of harmony and melody in rock music. Journal of New Music Research. 2(3), 187-204. Wald, E. (2009). How the Beatles Destroyed Rock 'n' Roll: An Alternative History of American Popular Music. Oxford: Oxford University Press. APPENDIX Songs Featured in the 1958-1971 Sub-Corpus Year Artist Title 1958 Count Basie Going to Chicago 1958 Chuck Berry Sweet Little Rock And Roll 1958 Johnny Cash The Ways of a Woman 1958 Jimmy Clanton Just A Dream 1958 The Everly Brothers Bird Song 1958 Peggy Lee Fever 1958 Louis Prima & Keely Smith That Old Black Magic 226 Empirical Musicology Review Vol. 10, No. 3, 2015 1959 LaVern Baker I Cried a Tear 1959 Rod Bernard This Should Go On Forever 1959 Chuck Berry Almost Grown 1959 Johnny Cash I Got Stripes 1959 Ray Charles (Night Time is) The Right Time 1959 Fats Domino Be My Guest 1959 Fats Domino I Want to Walk You Home 1959 Johnny Horton The Battle of New Orleans 1959 Johnny Horton Johnny Reb 1960 Brenda Lee Sweet Nothin’s 1959 Elvis Presley My Wish Came True 1959 Elvis Presley One Night 1959 Cliff Richard Living Doll 1959 Santo & Johnny Sleep Walk 1959 Dinah Washington Unforgettable 1960 Floyd Cramer Last Date 1960 Bing Crosby Silent Night 1960 Dion Where or When 1960 Jimmy Jones Handy Man 1960 Jimmy Reed Baby What You Want Me to Do 1960 Charlie Rich Lonely Weekends 1960 Marty Robins Big Iron 1960 The Ventures Perfidia 1961 Gary U.S. Bonds Quarter to Three 1961 James Brown Baby, You’re Right 1961 James Brown I Don’t Mind 1961 The Crystals (There’s No Other) Like My Baby 1961 Dion Runaround Sue 1961 Roy Drusky Three Hearts in a Tangle 227 Empirical Musicology Review Vol. 10, No. 3, 2015 1961 The Everly Brothers Ebony Eyes 1961 The Everly Brothers Walk Right Back 1961 Don Gibson Sea of Heartbreak 1961 Etta James Fool That I Am 1961 Ben E. King Amor 1961 Brenda Lee Dum Dum 1961 The Miracles Mighty Good Lovin’ 1961 Gene Pitney Town Without Pity 1961 Elvis Presley (Marie is the Name) His Latest Flame 1961 Elvis Presley Little Sister 1961 The Shirelles Will You Love Me Tomorrow? 1961 The String-A-Longs Wheels 1961 Johnny Tillotson Jimmy’s Girl 1961 Ike & Tina Turner It’s Gonna Work Out Fine 1962 Paul Anka Love Me Warm and Tender 1962 The Beach Boys Surfin’ Safari 1962 Bruce Chanel Hey! Baby 1962 Chubby Checker The Twist 1962 Nat “King” Cole Ramblin’ Rose 1962 The Contours Do You Love Me? 1962 Bing Crosby White Christmas 1962 The Crystals He’s a Rebel 1962 Bo Diddley You Can’t Judge a Book by the Cover 1962 Dion (I Was) Born to Cry 1962 Dion Love Came to Me 1962 Dion Lovers Who Wander 1962 The Falcons I Found a Love 228 Empirical Musicology Review Vol. 10, No. 3, 2015 1962 The Everly Brothers That’s Old Fashioned (That’s the Way Love Should Be) 1962 Eddie Holland Jamie 1962 Etta James Stop the Wedding 1962 Gladys Knight & The Pips Letter Full of Tears 1962 Brenda Lee All Alone Am I 1962 Brenda Lee Everybody Loves Me But You 1962 Brenda Lee Heart in Hand 1962 Little Joey & The Flips Bongo Stomp 1962 Jimmy Smith Walk on the Wild Side (Part 1) 1962 Hank Snow I’ve Been Everywhere 1962 Bertha Tillman Oh My Angel 1962 Johnny Tillotson It Keeps Right On A-Hurtin’ 1962 Mel Torme Comin’ Home Baby 1962 Dinah Washington Where Are You? 1963 LaVern Baker See See Rider 1963 Bobby Bare Detroit City 1963 The Beach Boys In My Room 1963 Solomon Burke If You Need Me 1963 The Chiffons He’s So Fine 1963 Nat “King” Cole Those Lazy-Hazy-Crazy Days of Summer 1963 Dion Ruby Baby 1963 Dion This Little Girl 1963 The Drifters On Broadway 1963 Dave Dudley Six Days on the Road 1963 The Fireballs Sugar Shack 229 Empirical Musicology Review Vol. 10, No. 3, 2015 1963 Etta James Would It Make Any Difference to You? 1963 Chris Kenner Land of the 1000 Dances 1963 Brenda Lee Loosing You 1963 Barbara Lewis Hello Stranger 1963 The Moments Walk Right In 1963 The Ronettes Be My Baby 1963 Johnny Tillotson Out of My Mind 1963 Johnny Tillotson Talk Back Trembling Lips 1963 Jackie Wilson Baby Workout 1964 Louis Armstong Hello Dolly! 1964 The Beach Boys Wendy 1964 The Beatles A Hard Day’s Night 1964 The Beatles Do You Want to Know a Secret? 1964 The Beatles I Saw Her Standing There 1964 The Beatles Love Me Do 1964 The Beatles She’s a Woman 1964 Jan & Dean Sidewalk Surfin’ 1964 Jan & Dean The Anaheim, Azusa & Cucamonga Sewing Circle, Book Review And Timing Association 1964 Jan & Dean The Little Old Lady (From Pasadena) 1964 Lesley Gore You Don’t Own Me 1964 B.B. King How Blue Can You Get? 1964 Brenda Lee As Usual 1964 Dean Martin Everybody Loves Somebody 230 Empirical Musicology Review Vol. 10, No. 3, 2015 1964 Willie Mitchell 20-75 1964 Elvis Presley Ask Me 1964 Ottis Redding Chained and Bound 1964 The Rolling Stones Not Fade Away 1964 The Rolling Stones Time Is On My Side 1964 Barbara Streisand Funny Girl 1964 Barbara Streisand People 1964 Irma Thomas Wish Someone Would Care 1964 Johnny Tillotson I Rise, I Fall 1964 Johnny Tillotson Worried Guy 1964 J. Frank Wilson & The Cavaliers Last Kiss 1965 Joan Baez There But For Fortune 1965 Fontella Bass Rescue Me 1965 The Beatles Eight Days a Week 1965 The Beatles Help! 1965 The Beatles I Don’t Want to Spoil the Party 1965 James Brown I Got You (I Feel Good) 1965 The Castaways Liar, Liar 1965 Ray Charles Crying Time 1965 Roy Head Treat Her Right 1965 Brenda Lee Too Many Rivers 1965 Martha & The Vandellas Nowhere to Run 1965 Bobbi Martin I Love You So 1965 Dean Martin I Will 1965 Buck Owens I’ve Got a Tiger by the Tail 1965 Wilson Pickett In the Midnight Hour 231 Empirical Musicology Review Vol. 10, No. 3, 2015 1965 Otis Redding I’ve Been Loving You Too Long (To Stop Now) 1965 The Righteous Brothers Unchained Melody 1965 Simon & Garfunkel The Sounds of Silence 1965 Sonny & Cher Baby Don’t Go 1965 The Yardbirds Heart Full of Soul 1966 The Byrds Eight Miles High 1966 Ray Charles Let’s Go Get Stoned 1966 The Chiffons Sweet Talkin’ Guy 1966 Bobby Darin If I Were a Carpenter 1966 Donovan Sunshine Superman 1966 The Kinks ‘Till the End of the Day 1966 Brenda Lee Coming On Strong 1966 Roger Miller You Can’t Roller Skate in a Buffalo Herd 1966 Aaron Neville Tell It Like It Is 1966 Buck Owens Think of Me 1966 Paul Revere & The Raiders Kicks 1966 The Righteous Brothers (You’re My) Soul and Inspiration 1966 The Rolling Stones Have You Seen Your Mother, Baby, Standing In The Shadow? 1966 Simon & Garfunkel A Hazy Shade of Winter 1966 Nancy Sinatra These Boots Are Made For Walkin’ 1966 Swingin’ Medallions Double Shot (Of My Baby’s Love) 1966 The Temptations Ain’t Too Proud to Beg 232 Empirical Musicology Review Vol. 10, No. 3, 2015 1966 The Yardbirds Shape of Things 1967 James Brown Cold Sweat – Part 1 1967 The 5th Dimension Go Where You Wanna Go 1967 The Animals San Franciscan Nights 1967 The Buckinghams Kind of a Drag 1967 The Byrds Goin’ Back 1967 Arthur Conley Sweet Soul Musi 1967 The Electric Prunes I Had Too Much to Dream (Last Night) 1967 The Four Tops Standing in the Shadows of Love 1967 Aretha Franklin Chain of Fools 1967 Aretha Franklin I Never Loved a Man (The Way I Love You) 1967 Marvin Gaye & Tami Terrell If I Could Build My Whole World Around You 1967 Marvin Gaye & Kim Weston It Takes Two 1967 Lesley Gore California Nights 1967 The Hollies Carrie-Ann 1967 The Music Explosion Little Bit O’ Soul 1967 The Music Machine The People in Me 1967 Nitty Gritty Dirt Band Buy For Me the Rain 1967 Roy Orbison Cry Softly Lonely One 1967 Wilson Pickett I’m in Love 1967 Wilson Pickett Soul Dance Number Three 1967 Elvis Presley Judy 1967 The Rolling Stones Dandelion 233 Empirical Musicology Review Vol. 10, No. 3, 2015 1967 Jimmy Ruffin I’ve Passed This Way Before 1967 Sagittarius My World Fell 1967 Simon & Garfunkel Fakin’ It 1967 Sopwith Camel Hello Hello 1967 The Turtles Happy Together 1967 The Who Happy Jack 1968 The Amboy Dukes Journey to the Center of the Mind 1968 Blue Cheer Summertime Blues 1968 The Box Tops Cry Like a Baby 1968 Canned Heat On the Road Again 1968 Clarence Carter Too Weak to Fight 1968 Ray Charles Eleanor Rigby 1968 Joe Cocker With a Little Help From My Friend 1968 Cream Sunshine of Your Love 1968 Creedence Clearwater Revival I Put a Spell on You 1968 Dion Abraham, Martin and John 1968 Flatt & Scruggs Foggy Mountain Breakdown 1968 Max Frost & The Troopers Shape of Things to Come 1968 Iron Butterfly In-A-Gadda-Da-Vida 1968 Tommy James Mony Mony 1968 The Miracles I Second That Emotion 1968 Wilson Pickett I Found a True Love 1968 Elvis Presley Guitar Man 1968 Elvis Presley If I Can Dream 1968 The Rascals People Got to Be Free 234 Empirical Musicology Review Vol. 10, No. 3, 2015 1968 Otis Redding (Sittin’ On) The Dock of the Bay 1968 Marty Robbins I Walk Alone 1968 Simon & Garfunkel Mrs. Robinson 1968 Steppenwolf Born to be Wild 1968 The Temptations I Wish It Would Rain 1968 B.J. Thomas Hooked On a Feeling 1968 Jerry Jeff Walker Mr. Bojangles 1969 Badfinger Maybe Tomorrow 1969 The Beach Boys Bluebirds Over the Mountain 1969 The Beatles Come Together 1969 Brother Jack McDuff Them From Electric Surfboard 1969 Glen Campbell Galveston 1969 Glen Campbell Wichita Lineman 1969 Jimmy Cliff Wonderful World, Beautiful People 1969 The Cowsills Hair 1969 The Cowsills Silver Threads and Golden Needles 1969 Creedence Clearwater Revival Bad Moon Rising 1969 Crosby, Stills & Nash Judy Blue Eyes 1969 The Isley Brothers It’s Your Thing 1969 The Jacksons I Want You Back 1969 Tommy James Crystal Blue Persuasion 1969 Janis Joplin Kozmic Blues 1969 Little Anthony & The Imperials Out of Sight, Out of Mind 1969 Peggy Lee Is That All There Is? 235 Empirical Musicology Review Vol. 10, No. 3, 2015 1969 The Miracles Baby, Baby Don’t Cry 1969 Oliver Good Morning Starshine 1969 The Rolling Stones Honky Tonk Women 1969 Sly & The Family Stone Hot Fun In the Summertime 1969 Sly & The Family Stone I Want to Take You Higher 1969 The Who Pinball Wizard 1969 The Youngbloods Get Together 1970 James Brown Get Up (I Feel Like a) Sex Machine – Part 1 1970 Glen Campbell It’s Only Make Believe 1970 Canned Heat Let’s Work Together 1970 Clarence Carter Patches 1970 Chicago 25 or 6 to 4 1970 Chicago Does Anybody Really Know What Time It Is? 1970 Chicago Make Me Smile 1970 Crosby, Stills & Nash Teach Your Children 1970 B.B. King The Thrill Is Gone 1970 The Meters Look-Ka Py Py 1970 Elvis Presley I Really Don’t Want to Know 1970 Ray Price For the Good Times 1970 Rare Earth Get Ready 1970 Jimmy & David Ruffin Stand By Me 1970 Santana Evil Ways 1970 Simon & Garfunkel Cecilia 1970 Simon & Garfunkel El Condor Pasa 1970 Edwin Starr War 236 Empirical Musicology Review Vol. 10, No. 3, 2015 1970 The Supremes Stoned Love 1970 R. Dean Taylor Indiana Wants Me 1970 The Tee Set Ma Belle Amie 1970 Ike & Tina Turner I Want to Take You Higher 1971 The 5th Dimension Never My Love 1971 The 5th Dimension One Less Bell to Answer 1971 The Band Life Is a Carnival 1971 Bread If 1971 Cornelius Brothers & Sister Rose Treat Her Like a Lady 1971 The Doors Riders of the Storm 1971 Emerson, Lake & Palmer Lucky Man 1971 Five Man Electric Band Absolutely Right 1971 Isaac Hayes The Look of Love 1971 Jimi Hendrix Freedom 1971 Elton John Levon 1971 Tom Jones She’s a Lady 1971 Roberta Flack & Donny Hathaway You’ve Got a Friend 1971 Marvin Gaye Mercy Mercy me (The Ecology) 1971 Tom T. Hall The Year That Clayton Delaney Died 1971 Gladys Knight & The Pips If I Were Your Woman 1971 Kris Kristofferson Loving Her Was Easier 1971 The Miracles I Don’t Blame You at All 1971 Graham Nash Chicago 1971 Ocean Put Your Hand in the Hand 1971 The Osmonds One Bad Apple 237 Empirical Musicology Review Vol. 10, No. 3, 2015 1971 Wilson Pickett Don’t Knock My Love – Part 1 1971 Elvis Presley There Goes My Everything 1971 Jerry Reed Ko-Ko Joe 1971 The Rolling Stones Wild Horses 1971 Sonny & Cher All I Ever Need Is You 1971 Rod Stewart Maggie May 1971 James Taylor Country Road 1971 Ten Years After I’d Love to Change the World 1971 Bill Withers Ain’t No Sunshine 1971 Bobby Womack That’s the Way I Feel About Cha 1971 Stevie Wonder If You Really Love Me 238